PLINK- chromosome ID disappears after running association tests

Question

[melaniesmontes]

I am using PLINK to run some GWAS analyses working on a non-model organism with 4000+ super-contigs. I've managed to successfully create a map file from a vcf file using vcftools and a chromosome map that I supplied... here is what the map file looks like:

head full.map

supercont1.1004_pilon   10      0       3307
supercont1.1004_pilon   10      0       3310
supercont1.1004_pilon   10      0       3313
supercont1.1004_pilon   10      0       3330
supercont1.1004_pilon   10      0       3361
supercont1.1004_pilon   10      0       3362
supercont1.1004_pilon   10      0       3400
supercont1.1004_pilon   10      0       3416
supercont1.1004_pilon   10      0       3417
supercont1.1007_pilon   18      0       414

However, when I run association tests on the data in plink, all the chromosomes are listed as 0 in the result files:

head full.qassoc.adjusted

 CHR     SNP      UNADJ         GC       BONF       HOLM   SIDAK_SS   SIDAK_SD     FDR_BH     FDR_BY
   0   20271  2.451e-22  1.239e-21  1.154e-17  1.154e-17        INF        INF  5.771e-18  6.542e-17 
   0   20271  2.451e-22  1.239e-21  1.154e-17  1.154e-17        INF        INF  5.771e-18  6.542e-17 
   0   16411  8.034e-15  2.365e-14  3.784e-10  3.783e-10  3.765e-10  3.765e-10  1.261e-10   1.43e-09 
   0   35808  4.862e-14  9.941e-12   2.29e-09   2.29e-09   2.29e-09   2.29e-09   4.58e-10  5.192e-09 
   0   35808  4.862e-14   1.43e-13   2.29e-09   2.29e-09   2.29e-09   2.29e-09   4.58e-10  5.192e-09 
   0   79625   4.19e-12   1.43e-13  1.973e-07  1.973e-07  1.973e-07  1.973e-07  3.289e-08  3.729e-07 
   0  101661  7.266e-08  2.566e-07   0.003422   0.003422   0.003416   0.003416  0.0004889   0.005542 
   0   94063  6.586e-07   7.75e-06    0.03102    0.03102    0.03054    0.03054   0.003878    0.04396 
   0   79929  1.263e-06  3.647e-06    0.05947    0.05946    0.05774    0.05773    0.00585    0.06632

Is this because the program does not like to have more than 22 chromosomes? Or do I need to reformat the names? Is there a way to work around this?

I also noticed that I have multiple snps with the same ID. I don't know if this is normal (to group snps which are physically close to one another) or some problem with vcf tools which I used to make the files. This has the unfortunate consequence that I do not know exactly which of the SNPs/base pair positions are the ones with significant p-values in my association tests. Is it possible to recode the map file somehow to give them each random but unique identifiers?

Thanks in advance!

[chrchang523]

Are you using plink 1.07 or 1.9? 1.07 does not keep track of arbitrary contig names, but 1.9 should.

[WouterDeCoster]

I formatted post for readability using the 101010 button, try to do the same in the feature to properly structure file content.

[WouterDeCoster]

I guess you could chose more descriptive IDs in your vcf file, which would then be used here as well?

Answer 1

[christopher medway]

The first column is CHR and the second is SNPID. Have you got these the wrong way around?

I think that chromosome has to be numeric and, because yours is not ["supercont1.1004_pilon"], it is returning a null ('0') chromosome value. You are probably also correct that only a chromosome value between 1-22 (23-26 from Sex/MT) will be acceptable.